Bird Community Differences in Mature and Second Growth Garúa Forest in Machalilla National Park, Ecuador

ORNITOLOGIA NEOTROPICAL 16: 163–180, 2005 © The Neotropical Ornithological Society

BIRD COMMUNITY DIFFERENCES IN MATURE AND SECOND GROWTH GARÚA FOREST IN MACHALILLA NATIONAL PARK, ECUADOR

Becker, C. Dustin1 & Ana Ágreda2

1Earthwatch Institute, 3 Clock Tower Place-Suite 100, P.O. Box 75, Maynard, Massachusettes 01754, USA. 2CECIA (Corporación Ornitológica del Ecuador), Joaquín Tinajero E-305 y Jorge Drom, Quito, Ecuador.

Resumen. – Diferencias entre las comunidades de aves de bosque de garua maduro y secundario en el Parque Nacional Machalilla, Ecuador. – Las estrategias de conservación a nivel de los ecosiste- mas son más eficaces si consideran el grado de sensibilidad que presentan los organismos frente a la per- dida o el cambio estructural de su hábitat. Este estudio presenta información sobre las comunidades de aves del bosque húmedo de la región Tumbesina del Ecuador y, compara mediante el uso de redes de neblina y observación directa, la composición de aves presentes en bosque de garúa maduro y secundario dentro del Parque Nacional Machalilla (PNM). En este estudio registramos 159 especies de aves durante las épocas seca y lluviosa de 1999 & 2000. Realizamos un esfuerzo total de muestreo de 1950 h de capturas con redes y capturamos 729 individuos pertenecientes a 85 especies. Alrededor del 60% de las especies capturadas en redes de neblina dentro de uno u otro tipo de bosque no se hallaban compartidas. La riqueza de especies fue mayor en bosque secundario y se sugiere algunas razones autoecológicas: 1) presencia de especies dependientes de bosque maduro ausentes en bosque secundario, 2) especies insectívoras genera- listas prefieren bosque secundario, y 3) especies migratorias prefieren habitats marginales secundarios. Tres especies de aves endémicas y amenazadas, Hylocryptus erythrocephalus, Leptotila ochraceiventris y Lathrotriccus griseipectus fueron registradas frecuentemente en bosque secundario y no en bosque maduro. La riqueza de especies pertenecientes a los géneros de hormigueros fue mayor en bosque maduro, y estuvo relacionada con la abundancia de artrópodos. Diecisiete especies se registraron únicamente en bosque de garúa maduro, pero ninguna se halla dentro de alguna categoría de amenaza. Entre las mas importantes podemos destacar, Gymnopithys leucaspis, Sclerurus guatemalensis y Turdus daguae. La alta detectabilidad de estas especies dentro del bosque sugiere su uso como especies indicadoras de bosque de garúa, el mismo que se halla en peligro de extinción. Abstract. – Species monitoring for conservation planning is aided by understanding speciesí sensitivities to habitat change or loss. We studied bird communities in garúa forest, a type of low elevation tropical cloud forest in western Ecuador and Peru that is threatened by land conversion. We used mist-nets and strip counts to compare bird communities in second growth and mature garúa forest at Machalilla National Park, Ecuador. During the two wet and two dry seasons of 1999 and 2000, we recorded 159 bird species. In 1950 h of mist netting, we captured 729 individuals of 85 species. Over 60% of the species were unique to either second growth or mature forest. Species richness of birds was greater in second growth than in mature garúa forest for reasons apparently unrelated to abundance or diversity of food resources. Generalist insectivore species, dry forest endemic birds, and Neotropical migrants readily used second growth forest, while mature garúa forest species were absent or significantly less abundant. Three endemic and endangered species, Henna-hooded Foliage-gleaner (Hylocryptus erythrocephalus), Ochre-bellied Dove ______1Current address: P.O. Box 1248, Zuni, New Mexico 87327, USA. E-mail: [email protected]

163 BECKER & ÁGREDA

(Leptotila ochraceiventris), and Gray-breasted Flycatcher (Lathrotriccus griseipectus) were found in second growth more often than in mature forest. Antbird species richness and abundance was greater in mature garúa forest, than in second growth, possibly due to differential arthropod abundance in forest litter. Seventeen bird species were only encountered in mature garúa forest and none were threatened. Among them, Bicolored Antbird (Gimnopithys leucaspis), Scaly-throated Leaftosser (Sclerurus guatemalensis), and Dagua Thrush (Turdus daguae) were easy to detect by sight or song, making them good indicator species for this type of tropical forest. Birds detected by observation and listening were similar in both forest types making trail surveys less useful than mist netting for detecting differences between bird communities in the two forest types. Accepted 4 March 2005. Key words: Tropical forest conservation, garúa forest, indicator species, endemic birds, western Ecuador, secondary forest, mature forest.

INTRODUCTION forests that obtain moisture from clouds, these forests strip moisture from oceanic In addition to studying the last primary forest mists and fog (Becker 1999). Regional botani- patches in the Neotropics, ornithologists are cal experts refer to this type of vegetation as trying to determine the value of human-mod- garúa forest (Elao 1996). Garúa forests have ified lands and second growth habitats for physiognomically typical cloud forest vegeta- bird conservation (Petit & Petit 2002). Since tion (dense loads of vascular epiphytes and the 1950ís, 95% of the primary forest in west- bryophytes on tree branches and trunks) and ern Ecuador has been converted to pasture, occur as low as 400 m (Parker & Carr, 1992). crops, and second growth forest (Dodson & Described by Chapman (1926), Best & Gentry 1991). Elsewhere in the Neotropics, Kessler (1995), and Stattersfield et al. (1998), such rapid change in land cover caused major garúa forest in the Colonche Hills are domi- declines and extirpations of birds (Kattán et nated by tree genera such as Beilschmiedia, al. 1994, Stratford & Stouffer 1999, Estrada et Rheedia, Ocotea, Gleospermum and Quararibea al. 2000). With a baseline understanding of (Elao 1996). what species avoid second growth and prefer Garúa forests are used by a wide variety of mature tropical forest, we can better predict birds in western Ecuador, including 22 spe- the future composition of bird communities cies of hummingbirds, 79 regionally endemic in tropical forests, evaluate the success of res- species, and 15 species listed as threatened toration efforts, and possibly thwart further and near-threatened (Becker & Lopéz-Lanús extinctions via habitat preservation. In this 1997). Also, microclimate differences paper, we compare the composition of bird between leeward and windward slopes of communities in mature and second growth garúa forest can strongly define bird commu- “garúa” (fog and mist) forests in Machalilla nities at a local level (Becker 1999). Still, little National Park, in the Tumbesian Endemic is known about the community of birds in Bird Area (EBA) of Ecuador, an area of sig- mature garúa forest, relative to second growth nificant importance for 55 restricted-range garúa forest and no research on avian sensi- bird species (Stattersfield et al. 1998, Birdlife tivity to forest degradation has been con- International 2003). ducted in this type of tropical forest. In this Along isolated and patchy mountain study, we address three questions: 1) Does ridges of the Tumbesian region, a humid for- bird species composition in mature garúa forest similar to high elevation Andean cloud est differ substantially from that in second forest exists. However, unlike Andean cloud growth garúa forest? 2) What factors influ-

164 BIRD ASSEMBLAGES IN GARÚA FORESTS ence species richness in mature versus second species richness is greater where food and growth garúa forest? and, 3) What bird spe- cover resources (fruits, flowers, insects, and cies unique to or abundant in mature garúa woody understory) are most abundant and forest could be good indicators of this habitat variable. type? Finally, for conservation work, the pres- Species richness of birds might be ence of certain bird species can be useful indi- expected to be greater in mature than second- cators for evaluating forest condition (Karr ary garúa forest, because mature forests are 1982, Stotz et al. 1996). Good indicator spe- taller and have more vertical heterogeneity cies are unique, or common in a particular than second growth forests (Rice et al. 1984). habitat, are easily detected in that habitat, and The number of migrant and resident bird spe- show high sensitivity to degradation, fragmen- cies has been found to increase with vertical tation, or loss of a particular habitat (Stotz et foliage diversity (Greenberg 1996), and al. 1996). Avian indicators have been mature forests offer greater potential for described for forty-one principal habitats in niche diversification (Hutchinson 1959, the Neotropics (Stotz et al. 1996), but not for Schoener 1986). On the other hand, species garúa forests of the Tumbesian endemic bird richness has generally been found to be area (EBA). greater in successional habitats than in climax ones because of the variety of generalists spe- METHODS cies attracted there (Ambuel & Temple 1983, Warkentin et al. 1995, Grey et al. 1998). We Study area. Machalilla National Park (hereaf- predict that bird communities in both mature ter, Machalilla) is located in Manabi Prov., and second growth forest patches on Cerro Ecuador, and encompasses an area of 55,095 San Sebastián will both be dominated by ha including La Plata Island and the ocean wide-spread generalist bird species, because surrounding it. Machalilla was established to the mature forest patches are relatively small. preserve coastal tropical dry forest, a threat- We also predict that second growth will have ened biome with high bird endemism (Stat- fewer cavity nesting and terrestrial insectivo- tersfield et al. 1998), and to protect important rous bird species, because they depend on marine resources (Zambrano & Vargas 1998). micro-habitats associated with old forest Below 300 m, Machalilla is covered by dry stands (Willis 1974, 1979; Leck 1979, Karr forest and arid scrub recovering from exces- 1982, Canaday 1996, Strattford & Stouffer sive grazing and removal of trees for charcoal 1999). production (Zambrano & Vargas 1998). Differences between bird species richness At elevations above 300 m, on west facing and assemblage composition in mature and slopes, most of Machalillaís primary garúa second growth forests have previously been forest was selectively harvested or completely explained by habitat selection based on: 1) cleared during the 1950ís and 60ís (Zambrano diversity of substrates, 2) abundance of food & Vargas 1998). Now, with protection from resources, 3) competitive species interactions, grazing and timbering, about 6000 ha of and 4) physical characteristics of the site Machalilla should eventually regenerate to (Wiens 1989). Since habitat selection by tropi- garúa forests (Cuéllar et al. 1992). Our study cal birds is often correlated with food of bird communities took place on Cerro San resources (Loiselle & Blake 1991) and micro- Sebastián, the highest peak in Machalilla (700 habitat features of nest sites (Borges & m) where approximately 400 ha of mature Stouffer 1999), we investigated whether bird garúa forest still remain and are surrounded

165 BECKER & ÁGREDA by second growth forest. highly mobile birds in the understory, and Average annual temperatures at Machalilla under-sample large-bodied birds and canopy are 18–22oC. Annual rainfall can vary from species (Remsen & Good 1996). To assess the none to 4000 mm, but averages 375 mm per degree of this bias in garúa forest, we aug- year (Cañadas-Cruz 1983). December and mented our list of birds in each forest type by January are normally the driest months, sampling along strip transects (Bibby et al. except during El Niño years when rainfall can 1992) where mist nets were located. Observ- be at least an order of magnitude greater than ers (A. Agreda, D. Becker, O. Rodriguez) average. Rainfall is highest during February to skilled at identifying birds by song and sight, May, peaking in March. Fog and mist are spent 2–3 h each morning and evening present from June to November and can con- recording bird species in each forest type. tribute as much water to coastal hill forest Effort per forest type was about 320 person- watersheds as annual rainfall (Becker 1999). h. Number of captures and detections were used to rank birds as common, uncommon or Bird community sampling. We used mist nets and rare in the two forest types, and to assess the observations along trails to sample birds in set of bird species detected by each method. mature and secondary garúa forests during We considered species “common” and the dry and wet seasons of 1999 and 2000. An “abundant” if they were seen, heard, or netted independent mist netting sample consisted of on most days of the study, “uncommon” if operating 10 mist nets (12 m x 2.8 m, 36 mm seen, heard, or netted on several sampling mesh) from 06:00 to 11:00 h for 3 consecu- days, and “rare” if they were seen, heard, or tive days in each forest type (mature, second- netted only once or twice during the study. ary). We conducted six independent samples of bird composition in mature forest and Vegetation sampling. In March (wet season), we seven in second growth forest. sampled forest vegetation in 10 x 12 m plots We placed mist nets on or adjacent to the placed at five randomly selected mist net loca- trail every 25–40 m starting with a randomly tions in each forest type in 1999. Percentage selected location along a trail. Independent ground and canopy cover were estimated sets of nets were separated by at least 500 m. visually by the same observer in each plot and Elevation, terrain, orientation, and slope were forest type. The species richness of trees, similar for the nets in both forest types. To shrubs, and herbs was determined by count- reduce edge effects (Restrepo & Gomez ing the number of different species of each 1998), nets in mature, closed-canopy forests plant type. Presence or absence of fruits and were at least 1000 m from second growth or flowers on trees and shrubs were noted and edge habitats. Likewise, nets in second growth estimates of abundance were made. We forest were at least 1000 m from mature for- recorded diameter at breast height (dbh) for est. Dry forest and scrub associated with trees (single woody stems with a dbh greater lower elevations (below 300 m) was about 5 than 10 cm), and estimated tree height to the km from garúa forest sampling areas. During nearest meter using hypsometers. Plants with a sampling session we weighed, made stan- single or multiple woody stems with a dbh dard morphometric measurements, banded, greater than 2.5 cm, but less than 10 cm, were identified to species, sexed, and aged (when recorded as “shrubs or saplings” and woody possible) each bird captured in the nets stems smaller than 2.5 cm in diameter were (Bibby et al. 1992; Ralph et al. 1993). counted as “seedlings”. In the center of each Mist-netting tends to over-sample small, plot, soil moisture was measured using a

166 BIRD ASSEMBLAGES IN GARÚA FORESTS

Kelway soil tester. In March 2000, vegetation three randomly selected mist-net sites and on data were collected in two 20 x 20 m plots two large vegetation plots (N = 5 per forest located in randomly selected, but representa- type). Arthropod samples were pooled by tive areas of each forest type. The sampling sampling technique, forest type, and season, method changed when project leadership and preserved in 70% alcohol. Pooled sam- shifted from Becker to Agreda. The large ples were classified to the lowest taxonomic plots were sub-divided into four quadrats of level possible (usually family or genus) by 10 x 10 m, and one of these was randomly entomologist, T. McKay (Kansas State Uni- selected to describe the understory layer, can- versity). opy cover, number and species of seedlings, height and dbh of trees and percentage cover Taxonomy, conservation status, and statistical analy- by shrubs and herbs. ses. We followed Ridgely & Greenfield (2001) for taxonomy and general ecology of bird spe- Arthropod sampling. To determine the relative cies in Ecuador, and Stattersfield et al. (1998) abundance and types of arthropods in the two for distribution and ecology of restricted- forest habitats we used sweep sampling range species in the Tumbesian EBA. A spe- (Southwood 1978 fide Borror et al. 1989) and cies was considered a generalist if it was developed a new leaf litter sampling method. observed or captured in several types of habi- Arthropods in leaf litter were quickly gathered tats, filled a known broad dietary niche (like into dip nets from a 50 x 50 cm frame placed some frugivores), and had no obvious mor- on the ground 5 m from the midpoint of each phological adaptation or specialized strategy mist net (ordered sampling with a random for obtaining food (Ridgely & Tudor 1989 & start). Arthropods found inside the frame 1994, Warkentin et al. 1995). Specialist guilds were classified to order, counted, and pre- included, cavity nesting, ant-following, bark- served in alcohol. Sixty independent leaf-litter probing, nectar-feeding, and leaf tossing. We samples were obtained from each forest type: used recent status categories for endanger- 40 in January 2000 (dry season) and 20 in ment and conservation concern published by March 2000 (wet season). Depth of the litter Birdlife International (2003). layer (cm) was measured with a ruler, and the Univariate and factoral analysis of variance air-dried litter (sticks, leaves, and organic (ANOVA), t-tests, and non-parametric statis- material above the soil) was weighed with a tical tests of data were performed using Stat- spring balance. Prior to collecting litter and view 4.5 (Roth et al. 1995). Alpha was set arthropods from the framed area, the percent- at 0.05 for all inferential statistics, and Fisherís age of bare ground, green vegetation, and protracted least significant difference (PLSD) decayed organic matter covering the surface were applied in post hoc tests to reduce Type 1 of each 2500 cm2 leaf-litter plot was estimated. errors associated with multiple comparisons. We sampled arthropods from surfaces of To compare species richness and diversity understory vegetation (shrubs, herbs, treelets, of bird communities associated with second tree trunks) using a 30-cm diameter sweep growth versus mature garúa forest, we used net. An independent sample consisted of 30 the software program EstimateS 5.0.1 (Col- sweeps, moving the net across vegetation to well 1997). We compared three estimators of the left and right of oneís body while advanc- species richness: CHAO 1, ICE, and ACE. ing slowly through the understory of a ran- CHAO 1 is a non-parametric estimator of domly selected 10 x 10 m area. Three sweep species richness based on patterns of abun- samples were completed in each forest type at dance (Chao & Lee 1992). ICE (Incidence-

167 BECKER & ÁGREDA

TABLE 1. Comparison of forest variables sampled in mature and secondary forests on Cerro San Sebastián, Machalilla National Park, Ecuador (Mean ± SD) during wet season 1999.

Variables Mature Secondary t-test P N (120 m2 plots) 5 5 % ground cover 63 ± 18 76 ± 17 2.21 0.03 % canopy cover 75 ± 15 46 ± 23 4.75 0.0001 Tree DBH (cm) 22 ± 19 12 ± 5 3.49 0.0008 Tree height (m) 13 ± 7 7 ± 3 4.87 0.0001 % shrub cover 30 ± 13 49 ± 9 2.78 0.024 % herb cover 70 ± 13 51 ± 9 2.70 0.024 Tree density1 25 ± 1.4 (SE) 16 ± 2.8 (SE) 4.0 0.05

1Data from 20 x 20 m plots. based coverage estimator) and ACE (Abun- were 20 to 25 m tall and canopy cover ranged dance-based coverage estimator) use proba- from 60 to 80% (Table 1). Understory cover bilities associated with finding rare species in ranged from 30 to 80% by herbaceous and the samples to derive estimates of total spe- woody plants in the Araceae, Cyclanthaceae, cies richness in a sampling area (Colwell & Gesneriaceae, Melastomataceae, and Rubi- Coddington 1994, Colwell 1997). aceae families. Mature garúa forest had 5 Using S-Plus 2000 (1999) mathematical times more canopy trees (> 15 m tall) than programming software, we calculated species second growth forest, which averaged only rarefaction curves (Magurran 1988) to com- two canopy trees per plot (166 per ha). Sec- pare species richness in the two forest types. ond growth garúa forest also had a lower can- We did Monte Carlo simulations with 1000 opy height (trees ranging from 5 to 10 m), less repetitions for equal sample sizes drawn ran- canopy cover, and thicker ground cover than domly from first capture data using subsets of mature forest (Table 1). Common tree and 50 individuals. We calculated 95% confidence shrub genera were different between forest intervals for the most species rich sample as types. Inga and Cordia were the most common an indication of the maximum variation tree genera in second growth and genera of around the rarefaction curves. Jaccard simila- Piperaceae shrubs dominated the understory rity indices and Morisita-Horn beta-diversity providing abundant vertical and horizontal were also compared for the avian communi- surfaces for birds. Both types of forest had ties sampled in the two forest types (Magu- similar amounts of genera of bamboo and rran 1988). bird of paradise (Heliconia spp.). Mean percentages of shrub and ground RESULTS cover were greater in second growth than in mature forest (Table 1). Species richness of Vegetation in garúa forest. Tree genera in mature trees (9 to 18 species per plot) and shrubs (13 forest at Cerro San Sebastián included Mora- to 16 species per plot) were not significantly ceae (Ficus, Paulsenia, and Cecropia), Lauraceae different in the two forest types. Herbs were (Ocotea and Cinnamomun), Bombacaceae (Mati- more diverse in mature forest than in the sec- sia), and Meliaceae (Trichillia). Canopy trees ond growth forest. Mature forest had a mean

168 BIRD ASSEMBLAGES IN GARÚA FORESTS

TABLE 2. Mean counts ± SD of arthropods in mature and secondary forest quadrants (area = 0.25 m2) at Cerro San Sebastián, Machalilla National Park, Ecuador, and statistical outcomes of Fisher’s PLSD post-hoc comparisons of means based on analysis of variance for each taxon by season (PS) and forest type (PF).

Arthropod taxa Mature forest Second growth PS PF Dry Wet Dry Wet Araneida (spiders) 6.0 ± 0.6 11.0 ± 1.3 4.7 ± 4.0 7.6 ± 4.0 0.000 0.008 Formicidae (ants) 6.0 ± 10.0 12.0 ± 18.0 1.6 ± 6.0 8.7 ± 11.0 0.003 0.05 Coleoptera (beetles) 0.6 ± 0.8 5.5 ± 6.0 0.2 ± 0.5 5.1 ± 4.0 0.000 ns Orthoptera (hoppers) 0.5 ± 0.7 0.5 ± 0.8 0.3 ± 0.6 0.05 ± 0.2 ns 0.06 Centi/millipedes 0.5 ± 1.0 3.0 ± 3.0 0 0.6 ± 0.9 0.000 0.000 Total count 18.1 ± 18.0 37.0 ± 31.0 7.5 ± 8.8 25.2 ± 16.0 0.0001 0.0024 of 7 ± 1.6 species of herbs in small quadrat lar, averaging 212 g per 50 x 50 cm quadrat plots, whereas second growth only averaged sample, but litter was significantly deeper 3.3 ± 3 species per plot (t7 = 2.7, P < 0.05). (Mean ± SD) in the mature forest, 41 ± 3 cm deep (n = 59) versus 29 ± 2 cm (n = 60) in

Resource abundance and richness in mature vs second the secondary forest (t117 = 3.4, P < 0.05). The growth forests. Abundance of food was either proportion of roaches and spiders was greater the same in both forest types or more abun- in dry conditions, while ants and isopods were dant in mature forest. Fruit counts were simi- more abundant in relatively moist conditions. lar in both forest types (Kruskal-Wallis; P = During the dry season of 2000, ants made up 0.61) across seasons. Flowers (mean ± SE) 22% of the mature forest sample, but were in greater abundance in the understory represented only 3% of the arthropod in sec- of mature forest (11± 1) flowers per 120 m2, n ond growth. During the wet season of 2000, = 5) than in second-growth (1.4 ± 2 flowers ants represented 54% and 45% of the arthro- per 120 m2, n = 5, P < 0.03) during the dry pods in mature and secondary forest, respec- season of 1999, but no significant difference tively. in flower abundance was detected during A similar number of arthropods per sweep other sampling periods. sample were recorded in both habitats with Arthropod abundance in litter varied by spiders, chrisomelid beetles, and lauxaniid season (P < 0.001) and forest type (P < flies most common. Litter samples from the 0.002). Mean counts of arthropods differed two forest types were dominated by arachnids according to moisture levels and litter depth (spiders), isopods (pillbugs), ants, and Blattaria associated with season and forest habitat type (roaches). Both forest types were similar in (Table 2). In mature forest, the mean number richness of arthropod taxa, averaging 20 fami- of arthropods ± SE was 25 ± 3 per litter sam- lies per sample for pooled sweeps and litter ple, nearly doubling the mean of 13 ± 2 in samples. The richness of non-insect taxa second growth forest sites (ANOVA, Fisherís (Chilopoda, Diplopoda, Isopoda, Araneida,

PLSD1,118; P < 0.003). During the dry season Acarina, Scorpionida, Chelonethida, Phalang- of 2000, soil moisture averaged 46 ± 12% in ida) averaged one taxon in sweep samples and the mature forest, but was only 12 ± 2% in five in litter samples (Fisherís PLSD1,7; P < second-growth forest (U16 = 13; P < 0.05). 0.02), but richness of arthropod taxa was not Biomass of litter in the two habitats was simi- significantly different by forest type.

169 BECKER & ÁGREDA

TABLE 3. Results of multivariate analysis of vari- between the two forest types. ance on daily captures of birds by forest type The mean number of birds netted per day (mature and second growth), season (dry and wet), varied by forest type, year, season, and day of year (1999 or 2000), and day (1, 2, 3). Data are a netting session (ANOVA, Table 3, Fig. 1). from mist-net samples on Cerro San Sebastián in Because “day” did not interact significantly Machalilla National Park, Ecuador. Insignificant with other variables, daily means were treated interactions are not shown. as independent in parametric statistical comparisons. In 1999, mean daily captures were Sources Number of birds captures significantly higher in second growth than in per day mature garúa forest, and higher during dry df FPthan wet season (ANOVA with post hoc Fish- Foret type 1,12 9.9 0.008 erís PLSD; P < 0.05, Table 3, Fig. 1). In 2000, Season 1,12 5.9 0.03 no significant differences in mean daily cap- Year 1,12 15.2 0.002 tures were noted by sample, season, or forest Day 2,12 7.0 0.009 type (Fig. 1). Overall, the mean number of species captured per day in second growth (15 Forest type * Year 1,12 8.3 0.01 ± 4.7) was greater than in mature garúa forest

(11.6 ± 3.4, t34 = -2.5, P = 0.02) and daily spe- Avian community: metrics and estimates of species cies richness was correlated with the number richness. We detected 159 bird species in Cerro of individuals captured in both habitats (r2 = San Sebastián study area (Appendix 1). Dur- 0.8, P < 0.001). ing 1950 h of mist netting, we captured 729 Rarefaction (Smith et al. 1985) and Monte individuals (866 captures in total) belonging Carlo simulation show that species accumula- to 85 species (53.4% of the total species tion in second growth consistently exceeds recorded). The percentage of recaptures that for mature forest (Fig. 2). Bird species across seasons was 15.8%. richness estimates generated by EstimateS An equal effort comparison using six were also higher for second growth than samples from each forest type resulted in 771 mature forest. ACE, ICE, and CHAO 1 esti- captures with more captures in second mates of species richness were 82 species ± growth: 56% vs 44% in the mature forest. 2.5, 84 ± 1.7, and 91 ± 15, respectively, for Forty-nine species were netted in mature second growth, but only 65 species ± 4, 63 ± garúa forest, whereas 66 species were netted 4.5, and 62 ± 8.5, respectively, for mature for- in second growth. The two forest types were est. estimated to share 42 species, a Morisita- Horn value of 0.58. When all mist-netting Composition of Avian Communities. Generalists results were included (uneven effort), a Jac- bird species dominated second growth areas, card similarity score (Magurran 1988) of 0.60 but were less abundant in the small patch of resulted from 32 species netted in both forest mature forest. Species richness of nectar feed- types (76% of the Morisita-Horn estimate). ers and frugivorous cavity nesting birds were Still, 53 species detected with mist nets were similar in the two forest types. Sallying insecti- unique to either mature or second growth vores and canopy insectivores (generalist garúa forest. Birds detected by observation insectivores) were more numerous in second were typically larger and were found in both growth (Fig. 3). Nearctic migrants, flexible forest types making strip counts less useful generalists in foraging and habitat use, repre- than mist netting for detecting differences sented 2.3% of the individuals captured in

170 BIRD ASSEMBLAGES IN GARÚA FORESTS

FIG. 1. Variation in mean capture rate of birds in mature and secondary garúa forest by season (dry, wet) and year on Cerro San Sebastián in Machalilla National Park, Ecuador. Error bars are standard deviations. second growth, showing a clear preference for growth. Ecuadorian Thrush (Turdus maculiros- marginal habitats (Fig. 3). Specialist insecti- tris) and Swainsonís Thrush (Catharus ustulatus) vores such as ant-following and leaf-tossing were fairly common in second growth, but terrestrial foragers, and bark-probers were absent in mature forest samples (Appendix 1, more prevalent in mature forest than second Table 3). Olivaceous Woodcreeper (Sittasomus 2 growth (Fig. 3; χ 8 = 67.6, P < 0.0001). griseicapillus) was common in second-growth, Of the 15 most common species sampled but was half as likely to be caught in mature in mist-nets, less than half were frequently forest (Appendix 1, Table 3). mist-netted in both forest types (Table 3). Moderately common species, representing Plain-brown Woodcreeper (Dendrocincla fuligi- at least 1% of the captures in both habitats, nosa) and Wedge-billed Woodcreeper (Glypho- accounted for 445 captures, or 51% of the rynchus spirurus) were abundant in mature total (Appendix 1). Nine of these 38 moder- forest samples, but were rare or absent in sec- ately common species had significantly higher ond growth samples (Appendix 1, Table 3). captures in mature forest than in the second Slaty Antwren (Myrmotherula schisticolor) and growth (P < 0.05, PLSD adjusted). For exam- Black-headed Antthrush (Formicarius nigricapil- ple, Spotted Woodcreeper (Xiphorhynchus eryth- lus) were among the 15 most commonly mist- ropygius), Plain-brown Woodcreeper (Dendro- netted birds in mature forest, but were not cincla fuliginosa), Chestnut-capped Brush-finch recorded in second growth. Captures of spot- (Buarremon brunneinucha), and Gray-breasted ted Nightingale-thrush (Catharus dryas) were Wood-Wren (Henicorhina leucophrys) were twice ten-fold higher in mature forest than second as likely to be captured in mature forest as in

171 BECKER & ÁGREDA

FIG 2. Rarefaction curves for equal-effort samples of bird species netted in mature and secondary garúa forest at Cerro San Sebastián, Machalilla National Park, Ecuador. The dotted lines are 95% confidence intervals. second growth. Woodcreeper (Dendrocolaptes certhia) are all cav- Four species had a statistically higher cap- ity nesters that require large and old trees for ture rate in second growth than in mature for- nesting (Ridgely & Greenfield 2001). Scaly- est: endemic Gray-and-gold Warbler throated Foliage-gleaner (Anabacerthia variegati- (Basileuterus fraseri), endemic and vulnerable ceps) forage on epiphytes, and these plants are Gray-breasted Flycatcher (Lathrotriccus grisei- more abundant on larger, older trees (C. D. pectus), Speckled Hummingbird (Adelomyia mel- Becker pers. observ.). Scaly-throated Leaf- anogenys), and Plain Antvireo (Dysithamnus men- tosser (Sclerurus guatemalensis), Dagua Thrush talis) (P < 0.05, PLSD adjusted). Of the 85 (Turdus daguae), and Orange-billed Sparrow species detected by mist nets, 17 species were (Arremon aurantiirostris) forage in moist litter unique to mature garúa forest (Appendix 1). on the mature forest floor, where litter is Of these 17 species, 13 are dependent on thicker, softer, and contains more arthropod mature or old-growth forest morphology for food resources in mature forest than in sec- their survival and reproduction. For example, ond growth forest. Finally, Immaculate Ant- Red-headed Barbet (Eubucco bourcierii), Red- bird (Myrmeciza immaculata) and Bicolored rumped Woodpecker (Veniliornis kirkii), Antbird (Gymnopithys leucaspis) are obligate Crimpson-rumped Toucanet (Aulacorhynchus army ant followers, and army ants were more haematopygus), Collared Trogon (Trogon collaris), available year round in mature than in second- Red-billed Scythebill (Campylorhamphus trochi- ary forest sites. lirostris), Wedge-billed Woodcreeper (Glypho- In total, 35 bird species were unique to rynchus spirurus), and Northern Barred- second growth (Appendix 1). Nineteen are

172 BIRD ASSEMBLAGES IN GARÚA FORESTS

FIG 3. Percentage of netted individuals in different trophic guilds in mature and secondary garœa forest on Cerro San Sebasti‡n, Machallia National Park, Ecuador. Asterisk indicates a significant difference in proportional representation of the guild by forest type (χ2 ; P < 0.05). widespread species commonly found in sec- row (Arremon abeillei). Nearctic migrants were ondary and scrub forest of northern South netted in second growth mainly during the America and parts of Central America (details wet season. Swainsonís Thrush were most in Ridgely & Greenfield 2001). Six species abundant followed by Acadian Flycatcher recorded only in second growth included (Empidonax virescens), and Summer Tanager Tumbesian endemics that mainly inhabit low- (Piranga rubra), and all were noted during both land dry forests and scrub: Henna-hooded years. One Olive-sided Flycatcher (Contopus Foliage-gleaner (Hylocryptus erytrhocephalus), cooperi) netted in second growth in 1999 was a Watkinís Antpitta (Grallaria watkinsi), Ecua- new record for the area. Other new records at dorian Thrush, Plumbeous-backed Thrush Machalilla were: Yellow-margined Flatbill (Tol- (Turdus reevei), Speckle-breasted Wren momyias flavotectus), Pacific Flatbill (Rhynchocy- (Thryothorus sclateri), and Black-capped Spar- clus pacificus), Song Wren (Cyphorhinus

173 BECKER & ÁGREDA phaeocephalus), and Smoke-colored Pewee (Con- response to forest degradation (Rappole & topus nigrescens). Morton 1985, Greenberg 1996, Borges & Stouffer 1999) and fragmentation (Bierre- Endangered and endemic birds. Several endan- gaard 1986, Parker & Carr 1992, Stouffer & gered and restricted-range species were found Bierrregaard 1995, Gascon-Claude et al. in higher numbers in second growth than in 1999). Community structure often changes in mature garúa forest. endemic and vulnerable response to disturbance (Connell 1978, Karr Gray-breasted Flycatcher was four times 1981, Tilman & Pacala 1993) and when moni- more likely to be captured in second growth tored over time, species richness, evenness, than in mature forest (Appendix 1). Four vul- and abundance are useful measures of com- nerable species, Rufous-headed Chachalaca munity resilience and stability (Magurran (Ortalis erythroptera), Little Woodstar (Chaetocer- 1988). At San Sebastián, avian species compo- cus bombus), Henna-hooded Foliage-gleaner, sition in mature and second growth garúa for- and Ochre-bellied Dove (Leptotila ochraceiven- est differed more than expected given the tris) were netted, heard, or observed in sec- small patch size of the mature forest. How- ond-growth, but were not detected by any ever, the two types of garúa forest differed method in mature forest (Appendix 1). substantially accentuating the importance of Ochraceous Atilla (Attila torridus), vulnerable, habitat variables for the presence and absence was commonly heard in the canopy of forest of bird species. Tree and shrub species were borders as well as inside mature forest at San entirely different in the two forest types. Sebastián. Gray-backed Hawk (Leucopternis Mature forest had old, large trees with abun- occidentalis), globally endangered, was heard dant epiphytes, and had more invertebrate several times at Cerro San Sebastián in both food resources in a deep litter layer. Second years, but showed no obvious associations by growth forest had more horizontal and verti- forest type. Capture rates of other endemic cal woody structure in the understory, species are listed in Appendix 1. Given the loss of primary forest and the rate of fragmentation of mature garúa forest Strip counts. There were no significant differ- patches in western Ecuador, the sustainability ences in species abundance in mature vs sec- of such isolated and distinctive bird commu- ond growth forest based on strip count nities is probably in peril. We suggest that results. Strip counts were more effective than staff at Machalilla monitor the bird commu- mist nets for detection of large and/or vocal nity in mature garúa forest, using the most birds like tinamous, toucans, ant-thrushes, sensitive indicator species described in this parrots, and raptors. Appendix 1 provides a study, and that they promote the regeneration list of species detected during strip counts of mature forest through reforestation and and their categories of abundance (common, removal of grazing animals in the highlands. uncommon, rare). Indicator species for mature Garúa forest. Because they are relatively common, easily DISCUSSION detected throughout the year with minimal netting, and appear to show high sensitivity to Tropical bird assemblages change along natu- habitat degradation, at least 23 bird species ral gradients of elevation (Terborgh & Weske serve as good indicator species for mature 1975, Young et al. 1998, Blake & Loiselle garúa forest (see Appendix 1). Thirteen of 2000) and at ecotones (Smith 1977). Substan- these species are especially useful as they tial divergence in bird assemblages occur in exhibit behavioral adaptations for mature for-

174 BIRD ASSEMBLAGES IN GARÚA FORESTS est such as nesting in older trees with cavities, between the western Andes and the coastal foraging on epiphytes, or requiring deep, hills. Furthermore, some garúa forest species moist leaf litter to obtain terrestrial arthro- have recently been recognized as new species pods. The presence and proportions of partic- [e.g., White-necked Thrush (Turdus albicollis) ular antbird species might also be used for recently split as Dagua Thrush (Turdus daguae) habitat evaluation in garúa forest. At Macha- in Ridgely & Greenfield (2001)]. As we learn lilla, Bicolored and Immaculate antbirds more about the genetics and ecology of garúa avoided second growth, whereas White- forest birds of the Tumbesian region, we may backed Fire-eye (Pyriglena leuconota) used realize that some species are more of a con- both forest types. In general, ant-following servation concern than previously thought. bird species are extinction prone (Willis 1978, 1979) and they respond negatively to roads Species richness in second growth forest. Relative to (Canaday 1996) and fragmentation (Stouffer mature forest, the high avian species richness & Bierregaard 1995) making their decline a in second growth appears to be caused by the potential indicator of negative anthropogenic absence of mature forest specialists, and the activities and their increase a potential sign of presence of widespread generalist insectivores forest habitat improvement. and canopy foragers. Lack of territorial spe- Avian indicators in Ecuadorian garúa for- cialists would permit an influx of species (Til- est show little overlap with other major Neo- man 1999) from different sources (low tropical habitats (Stotz et al. 1996). Our list of elevation dry forest, long-distance migratory indicator species for garúa forest shares only pathways, dispersal from other moist forest three species with the list for the Chocó low- areas) filling the vacant niches (Holt 1993). land forest of northwestern Ecuador (Stotz et Second growth forest on Cerro San Sebatián al. 1996): Immaculate Antbird, Barred Wood- was well-represented by generalist species creeper, and Red-billed Scythebill. Chocó low- from the lowlands, as 41.7% of birds captured land forest is the closest endemic bird area and observed in this forest type are common (EBA) to Machalilla National Park, and is inhabitants of lower elevation transitional and located immediately north of the Tumbesian dry forest. EBA. Generalists tend to replace specialists Given the unique assemblage of birds in when habitats are degraded (Ambuel & Tem- mature garúa forest and the rapid decline in ple 1983, Warkentin et al. 1995, Grey et al. this forest type, it is surprising that none of 1998). Second growth lacks the habitat fea- the birds we mist-netted are species of conser- tures of mature forest (tall canopy covered vation concern. Many bird species in mature with epiphytes, moist soil with a higher abun- garúa forest may belong to historically wide- dance of arthropods) so specialists like ant- spread populations now isolated due to frag- birds and leaf tossers avoid such areas, mentation. Plant communities on Cerro San creating vacancies for generalist species. Ant- Sebastián have greater similarity to those birds are known to rely on moist mature for- observed in Río Palenque, 200 km northeast est settings where large arthropods (Willis in the Andean foothills than with forest com- 1978) and bivouacs for ants (Roberts et al. munities of the Chocó (northwestern Ecua- 1998) are sustained. Since more species can dor) or Jauneche region (headwaters of rio coexist in marginal habitat if they are tempo- Guayas) (Parker & Carr 1992). Gentry (fide rary visitors in space and time (Nee & May Parker & Carr 1992) speculated that forest 1992), high turnover rates of species in sec- cover had previously been contiguous ond growth forest also help explain the more

175 BECKER & ÁGREDA species-rich assemblage there. Migrants from Good (1996) outline many pitfalls of interpre- North America, common during the dry sea- ting data based on mist netting only, it son, contributed to turnover of species in sec- remains a reliable method for sampling terres- ond growth. trial insectivores and small understory species. Avian species richness in garúa forest was We failed to detect major differences in com- not correlated with food resource diversity or munity composition between mature and abundance. For example, although litter second growth garúa forest using strip arthropods were more abundant in mature counts, but mist netting revealed major diffe- forest, avian species richness was lower there rences in bird composition. Our mist-netting than in second growth forest. According to effort sampled slightly more than half the sweep net samples, and flower and fruit bird species recorded in garúa forest habitats. counts, there were no striking differences in Most were small, highly mobile insectivorous food resources available to birds in the under- birds that use the ground and understory story vegetation of the two forest types, yet layer of the forest. Mist netting is an appro- avian diversity was greater in second growth priate sampling method for detecting avian forest. responses to degradation because so many One clear difference between the two for- sensitive species are terrestrial insectivores est types was shrub cover. Shrubby under- (Stouffer & Bierregaard 1995, Canaday 1996). growth in secondary forest provides foraging More studies combining mist netting and cen- sites and perches, possibly accommodating suses along transects would increase our more birds in the forest strata best sampled understanding of probabilities of consistent by mist-nets. Canopy birds typically follow detection of birds in different strata of the the foliage-air interface as they forage forest and biases of mist netting in the tro- (Schemske & Brokaw 1981). Canopy species pics. [Bay-headed Tanager (Tangara gyrola), Lesser Greenlet(Hylophilus decurtatus), Bananaquit Implications for conservation in the Tumbesian (Coereba flaveola), Tropical Parula (Parula endemic bird area. Some endemic bird species of pitiayumi), and non-migratory Red-eyed the Tumbesian region in southwestern Ecua- Vireo, (Vireo olivaceous chivi)] were observed dor are associated with specific habitat types in old-growth forest, but were more fre- (Stattersfield et al. 1998). Five species, Black- quently netted in second growth. ish-headed Spinetail (Synallaxis tithys), Pacific We cannot eliminate landscape level pro- Parrotlet (Forpus coelestis), Short-tailed Hum- cesses as an explanation for the different spe- mingbird (Myrmia micrura), Collared Antshrike cies counts in mature and second growth (Sakesphorus bernardi), and Watkinsí Antpitta garúa forest. Since mature forest is relatively are mainly found in scrub forest at low eleva- isolated and small in area, forest dependent tions (below 200 m). Scrub dry forest in west- species may have already been lost due to area ern Ecuador, a special vegetation type found effects (Stratford & Stouffer 1999). Compar- on the Pacific coast, is heavily disturbed by ing species composition in larger fragments human activities. This disturbance has a long of garúa forest along the Chongón-Colonche history, with evidence of human settlements Cordillera, where Cerro San Sebastián is dating 6000 B.C. (e.g.,Valdivia). This may be located, may help determine the effects of why some lowland species appear to be fragmentation. adapted to secondary habitats and disturbed forest. This highlights the importance of Biases in mist netting. Although Remsen & understanding variation in speciesí sensitivi-

176 BIRD ASSEMBLAGES IN GARÚA FORESTS ties to habitat change or loss in this region the region. Garúa forest in the Tumbesian and for gathering information on habitat region are unique and dynamic due to pro- requirements of endemic bird species in order nounced annual fluctuations in rainfall and to determine the degree of habitat specializa- fog capture. Bird populations seem exhibit tion of some of these species. some adaptations to climate fluctuation An important trait in the life history of including vertical migration and use of multi- many Tumbesian endemics might be their ple habitat types. ability to move across habitats and to disperse Social interactions, resource availability, or perform seasonal and altitudinal move- and autecological factors interact over short ments in different habitats. Evidence for local time scales, while vicariance, dispersal, and migration exists given the variety of species speciation operate over longer time scales to associated with specific habitats and eleva- determine assemblages of species in a particu- tions in different months of the year (Best lar place, season, and time (Wiens 1989, Rob- 1992). We have also noted seasonal move- inson et al. 1992). Given that many Tumbesian ments along altitudinal gradients and between endemic and threatened species use second dry and humid cloud forest at Loma Alta growth garúa forest, preservation of garúa Ecological Reserve (Becker, field data) for forest in various successional stages is an Henna-hooded Foliage-gleaner, Ochre-bel- important conservation goal for western lied Dove, Little Woodstar, Plumbeous Ecuador (Becker & Agreda 2001). Thrush, Gray-breasted Flycatcher, Gray-and- gold Warbler, and Black-capped Sparrow. All ACKNOWLEDGMENTS these species have been recorded in secondary forest at different elevations and may depend We thank Carlos Zambrano for permission to on natural corridors of humid forests, strips complete fieldwork in Machalilla National of forest along rivers or patches of transitional Park, and the Dirección de Áreas Naturales y forest at intermediate elevations, to move Vida Silvestre of Ministerio de Ambiente for between habitats. For this reason conserva- national level research permits. Earthwatch tion efforts should focus on the protection of Institute provided all of the funding for the elevational gradients in the Chongón- study. We are especially grateful to Earth- Colonche Cordillera. watch volunteers for their help collecting the Machalilla National Park is an important data with the best spirit possible under diffi- reserve because it encompasses an ecological cult conditions. We appreciate the help pro- gradient that includes plant communities of vided by local research assistants Néstor deciduous and humid cloud forests. This Merchán and Pascual Torres, and biologists study provides information on the differences Andrea Richter and Orfa Rodríguez. We in bird communities in mature and second thank K. Garrett for assistance with rarefac- growth humid forest and the use of second tion, T. McKay for insect classifications, and growth forest habitat by many Tumbesian D. Dobkin, M. Hollingsworth, B. Jensen, M. endemics. We identified specific bird species Johnson, A. Povilitis, B. Sandercock, and P. assemblage in mature and second growth Stouffer for constructive comments on the garúa forest and provide a list of species use- manuscript. ful as indicators of mature forest conditions. Such information is valuable for management REFERENCES of garúa forest in Machalilla, and for monitoring the expansion of mature garúa forest in Ambuel, B., & S. A. Temple. 1983. Area-dependent

177 BECKER & ÁGREDA

changes in bird communities and vegetation in ecológico del Ecuador. Ministerio de Agricul- southern Wisconsin forests. Ecology 64: 1057– tura y Ganadería, Quito, Ecuador. 1068. Canaday, C. 1996. Loss of insectivorous birds Becker, C. D., & B. López-Lanús. 1997. Conserva- along a gradient of human impact in Amazo- tion value of a garúa forest in the dry season: a nia. Biol. Conserv. 77: 63–77. bird survey in the Reserva Ecológica de Loma Chao, A., & S. M. Lee. 1992. Estimating the num- Alta, Ecuador. Cotinga 8: 66–74. ber of classes via sample coverage. J. Am. Stat. Becker, C. D. 1999. Protecting a garúa forest in Assoc. 87: 210–217. Ecuador: The role of institutions and ecosys- Chapman, F. M. 1926. The distribution of bird-life tem valuation. Ambio 28: 156–161. in Ecuador. Bulletin of the American Museum Becker, C. D., & A. E. Ágreda. 2001. Bird species of Natural History, Volume 55, New York, in the moist highlands of Machalilla National New York. Park. Field Report, CECIA. Quito, Ecuador. Collar, N. J., M. J. Crosby, & A. J. Stattersfield. Best, B. J. 1992. The threatened forests of south- 1994. Birds to watch 2: The world list of threat- west Ecuador. Biosphere Publications, Leeds, ened birds. BirdLife International, Conserva- U.K. tion Series No. 4, Smithsonian Institution Best, B. J., & C. T. Clarke. 1991. The threatened Press. Washington, D.C. birds of the Sozoranga region, southwest Ecua- Colwell, R. K. 1997. EstimateS 5.1. http://vice- dor. ICBP Study Report 44, Cambridge, U.K. roy.eeb.uconn.edu/Estimates_f/ Best, B. J., & M. Kessler. 1995. Biodiversity and Colwell, R. K., & J. A. Coddington. 1994. Estimat- conservation in Tumbesian Ecuador and Peru. ing terrestrial biodiversity through extrapola- BirdLife International, Cambridge. UK. tion. Philos. Trans. R. Soc. Lond. Ser. B 345: Bibby, C. J., N. D. Burgess, & D. A. Hill. 1992. Bird 101–118. census techniques. Academic Press, San Diego, Connell, J. H. 1978. Diversity in tropical rain for- California. ests and coral reefs. Science 199: 1302–1310. Bierregaard, R. O. 1986. Changes in bird commu- Cuéllar, J. C., M. Melo, J. C. Reyes, C. Landázuri, O. nities in virgin forest and isolated Amazonian Yépez, & J. Moreno. 1992. Estudio de las forest fragments. Ibis 128: 166–167. poblaciones del área interna y la zona de influ- BirdLife International. 2003. BirdLifeís online encia del Parque Nacional Machalilla. Fun- world bird database: the site for bird conserva- dación Natura, Quito, Ecuador. tion. Version 2.0. Cambridge, UK: BirdLife Diamond, J. 1974. Assembly of species communi- International. Available: http://www. ties. Pp. 342–444 in Cody, M. L., & J. M. Dia- birdlife.org mond (eds.). Ecology and evolution of Blake, J. G., & B. A. Loiselle. 2000. Diversity of communities. Belknap Press, Cambridge, Mas- birds along an elevational gradient in the Cor- sachusetts. dillera Central, Costa Rica. Auk 117: 663–686. Dodson, C. H., & A. H. Gentry. 1991. Biological Borges, S. H., & P. S. Stouffer. 1999. Bird commu- extinction in western Ecuador. Ann. Mo. Bot. nities in two types of anthropogenic succes- Gard. 78: 273–295. sional vegetation in central Amazonia. Condor Elao, C. B. 1996. Los árboles del Bosque Protector 101: 529–536. de Loma Alta. Jornadas de Biología, Univer- Borror, D. J., C. A. Triplehorn, & N. F. Johnson. sidad Técnica de Esmeraldas, Esmeraldas, 1989. An introduction to the study of insects. Ecuador. Saunders, Philadephia, Pennsylvania. Estrada, A., R. Cammarano, & R. Coates-Estrada. Brunnell, F. L. 1999. What habitat is an island? Pp. 2000. Bird species richness in vegetation fences 1–31 in Rochelle, J. A., L. A. Lehmann, & J. and strips of residual rain forest vegetation at Wisniewski (eds.). Forest wildlife and fragmen- Los Tuxtlas, Mexico. Biodivers. Conserv. 9: tation: management implications. Koninklijke 1399–1416. Brill NV, Leiden, Netherlands. Gascon, C., T. E. Lovejoy, R. O. Bierregaard, J. R. Cañadas, L. 1983. El mapa mioclmático y Malcolm, P. C. Stouffer, H. L. Vasconcelos, W.

178 BIRD ASSEMBLAGES IN GARÚA FORESTS

F. Laurance, B. Zimmerman, M. Tocher, & S. Washington, DC. Borges. 1999. Matrix habitat and species rich- Petit, L. J., & D. R. Petit. 2002. Evaluating the ness in tropical forest remnants. Biol. Conserv. importance of human-modified lands for neo- 91: 223–229. tropical bird conservation. Conserv. Biol. 17: Greenberg, R. 1996. Managed forest patches and 687–694. the diversity of birds in southern Mexico. Pp. Ralph, J. C., G. R. Geupel, P. Pyle, T. E. Martin, & 59–90 in Schelhas, J., & R. Greeberg (eds.). For- D. F. DeSante. 1993. Handbook of field meth- est patches in tropical landscapes. Island Press, ods for monitoring landbirds. USDA General Covelo, California. Technical Report PSW-GTR-144, Washington, Grey, M. J., M. F. Clarke, & R. H. Lyon. 1998. DC. Influences of the Noisy Miner Manorina melano- Rappole, J. H., & E. S. Morton. 1985. Effects of cephala on avian diversity and abundance in habitat alteration on a tropical avian forest remnant grey box woodland. Pac. Conserv. community. Pp. 67–95 in Buckley, M.S., R. Fos- Biol. 4: 55–69. ter, E. S. Morton, R. S. Ridgely, & F. G. Buckley Holt, R. D. 1993. Ecology at the mesoscale: the (eds.). Neotropical ornithology. American influence of regional processes on local com- Ornithologistsí Union, Washington, DC. munities. Pp. 77–88 in Ricklefs, R. E., & D. Remsen, J. V., & D. A. Good. 1996. Misuse of data Schluter (eds.). Species diversity in ecological from mist-net captures to assess relative abun- communities: historical and geographical per- dance in bird populations. Auk 113: 381–398. spectives. Univ. of Chicago Press, Chicago, Illi- Restrepo, C., & N. Gómez. 1998. Responses of nois. understory birds to anthropogenic edges in a Hutchinson, G. E. 1959. Homage to Santa Rosalia: neotropical montane forest. Ecol. Appl. 8: or, why are there so many kinds of animals. 170–183. Am. Nat. 95: 137–145. Rice, J., B. W. Anderson, & R.. D. Ohmart. 1984. Karr, J. R. 1981. Assessment of biotic integrity Comparison of the importance of different using fish communities. Fisheries 6: 21–27. habitat attributes of avian community organiza- Karr, J. R. 1982. Avian extinction on Barro Colo- tion. J. Wildl. Manage. 48: 895–911. rado Island, Panama: a reassessment. Am. Nat. Ridgely, R. S., & P. J. Greenfield. 2001. The birds of 119: 220–239. Ecuador. Volume I and II. Cornell Univ. Press, Kattán, G. H., H. Álverez-López, & M. Giraldo. Ithaca, New York. 1994. Forest fragmentation and bird extinc- Ridgely, R. S., & G. Tudor. 1989. The birds of tions: San Antonio eighty years later. Conserv. South America. Volume I. Univ. of Texas Press, Biol. 8: 138–146. Austin, Texas. Leck, C. F. 1979. Avian extinctions in an isolated Ridgely, R. S., & J.G. Tudor. 1994. The birds of tropical wet-forest reserve, Ecuador. Auk 96: South America. Volume II. Univ. of Texas 343–352. Press, Austin, Texas. Loiselle, B. A., & J. G. Blake. 1991. Temporal varia- Roberts, D. L., J. C. Robert, & L. J. Petit. 1998. Use tion in birds and fruits along an elevational gra- of premontane moist forest and shade coffee dient in Costa Rica. Ecology 72: 180–193. agroecosystems by army ants in western Pan- Magurran, A. E. 1988. Ecological diversity and its ama. Conserv. Biol. 14: 192–199. measurement. Princeton Univ. Press, Princeton, Robinson, G. R., R. D. Holt, M. S. Gaines, S. P. New Jersey. Hamburg, E. A. Marinko, & H. S. Fitch. 1992. Nee, S., & R. May. 1992. Dynamics of metapopula- Diverse and contrasting effects of habitat frag- tions: Habitat destruction and competitive mentation. Science 257: 524–26. coexistence. J. Anim. Ecol. 61: 37–40. Roth, J., K. Haycock, J. Gagnon, C. Soper, & T. Parker, T. A., III, & J. L. Carr. 1992. Status of forest Calrola. 1995. Statview. Abacus Concepts, Inc. remnants in the Cordillera de la Costa and adja- Berkeley, California. cent areas of southwestern Ecuador. Conserva- Schemske, D. W., & N. Brokaw. 1981. Treefalls and tion International, RAP Working Papers 2, the distribution of understory birds in a tropical

179 BECKER & ÁGREDA

forest. Ecology 62: 938–945. competition in the distribution of Andean Schoener, T. W. 1986. Resource partitioning. Pp. birds. Ecology 56: 562–576. 91–126 in Kikkawa, J., & D. J. Anderson (eds.). Tilman, D. 1999. Diversity by default. Science: Community ecology: pattern and process. 495–96. Blackwell Scientific Publications, Oxford, UK. Tilman, D. & S. Pacala. 1993. The maintenance of Smith, E. P., P. M. Steward, & J. Cairns. 1985. Simi- species richness in plant communities. Pp. 13– larities between rarefaction methods. Hydrobi- 25 in Ricklefs, R. E., & D. Schluter (eds.). Spe- ologicia 120: 167–170. cies diversity in ecological communities. Univ. Smith, T. B., R. K. Wayne, D. J. Girman, & M. W. of Chicago Press, Chicago, Illinois. Bruford. 1997. A role for ecotones in generat- Warkentin, I. G., R. Greenberg, & J. S. Ortiz. 1995. ing rainforest biodiversity. Science 276: 1855– Songbird use of gallery woodlands in recently 1857. cleared and older settled landscapes of selva Southwood, T. R. E. 1978. Ecological methods Lacandona, Chiapas, Mexico. Conserv. Biol. 9: with particular reference to the study of insect 1095–1106. populations. 2nd ed. Chapman and Hall, New Wiens, J. A. 1989. Spatial scaling in ecology. Funct. York, New York. Ecol. 3: 385–97. Stattersfield, A. J., M. J. Crosby, A. J. Long, & D. C. Willis, E. O. 1974. Populations and local extinc- Wege. 1998. Endemic bird areas of the world: tions of birds on Barro Colorado Island, Pan- priorities for conservation. Conservation Series ama. Ecol. Monogr. 44: 153–169. No. 7, Birdlife International, Cambridge, UK. Willis, E. O. 1978. Birds and army ants. Annu. Rev. Stotz, D. F., J. W. Fitzpatrick, T. A. Parker, & D. K Ecol. Syst. 9: 243–263. Moskovits. 1996. Neotropical birds: ecology Willis, E. O. 1979. The composition of avian com- and conservation. Univ. of Chicago Press, Chi- munities in remanescent woodlots in southern cago, Illinois. Brazil. Pap. Avulsos Zool. 33: 1–25. Stouffer, P. C., & R. O. Bierregaard. 1995. Use of Young, B. E., D. DeRosier, & G. V. N. Powell. Amazonian forest fragments by understory 1998. Diversity and conservation of understory insectivorous birds. Ecology 76: 2429–2445. birds in the Tilarán Mountains, Costa Rica. Auk Stratford, J. A., & P. C. Stouffer. 1999. Local 115: 998–1016. extinctions of terrestrial insectivorous birds in a Zambrano, C., & V. H. Vargas. 1998. Parque fragmented landscape near Manaus, Brazil. Nacional Machalilla. Ministerio de Ambiente. Conserv. Biol. 13: 1416–1423. Quito, Ecuador. Terborgh, J. W., & J. S. Weske. 1975. The role of

APPENDIX 1. Bird species mist-netted and observed in mature and second growth garúa forest on Cerro San Sebastián, Machalilla National Park, Ecuador.

This appendix is available on-line at: http:// www......

180